School Neighborhood Food Environments and Schoolchildren’s Nutrition in Greater Tunis.

Abstract

Background Over the last decades, childhood overweight and obesity have dramatically increased in low- and middle-income countries (LMICs), with the Middle East and North Africa (MENA) region experiencing one of the largest increases in childhood obesity rates. School neighborhood food environments include the multitude of food options available to children on their daily trajectories to/from school, and are thus considered a key contributor to children’s food choices and subsequent weight status. While there is considerable research from high-income countries on food environments, scarce evidence exists in low and middle-income countries (LMICs). This study aimed to fill this research gap by exploring school neighborhood food environments in Greater Tunis, Tunisia – a context typical of a highly developed and urbanized middle-income setting with an ongoing nutrition transition and increasing rates of childhood obesity. The objectives of this study were to: (1) document the obesogenicity of school neighborhood food environments using geospatial data, by measuring availability and accessibility of food retailers and advertisements in Greater Tunis; and (2) examine the impact of the retail food environment on children’s weight status in this same setting. It is the first study to do so in an LMIC of the MENA region.

Methods In this cross-sectional study conducted in 2020, a representative sample of 9 to 12-year-old schoolchildren in Greater Tunis were recruited from 50 schools. Geographic coordinates and pictures of all food retailers and advertisements, available within an 800-meter road-network buffer of primary schools, were collected through ground truthing using geospatial static methods. Retailers were classified into typologies (e.g. supermarket, ambulant vendor, pastry shop) and advertisements into food groups (e.g. fresh fruits and vegetables, processed meats, sugar-sweetened beverages). Then, they were all further categorized as healthy, mixed, or unhealthy according to an adapted NOVA classification system. Data at the child-, household-, and school-level were collected through questionnaires and included socio-demographic, dietary habits, and school-food policy variables, among others. Anthropometric measurements were taken and overweight and obesity were defined based on WHO growth standards. Count, density, and distance from schools to nearest retailers were assessed. Associations between school characteristics and types of food retailers or advertisement were explored by relative prevalence ratios (RPR) using crude and adjusted multinomial regression models. Association between availability (i.e., count and count/area) as well accessibility (i.e., distance from school to nearest retailer) of the different types of food retailers and children’s weight status was assessed using crude and adjusted multilevel logistic regressions to account for clustering of subjects within schools. A directed acyclic graph (DAG) was drawn to identify the set of confounding variables to be added in our models. Sensibility analyses were conducted using different metrics (counts and count/area) and weight-related outcomes (obesity, overweight, and BMI z-score) to assess the robustness of findings and support comparability across studies.

Results Overall, 3,621 food retailers and 2,098 advertisements were mapped. The median counts of food retailers and advertisements were 64/school (IQR=47-95) and 36/school (IQR=25-53), respectively. About two thirds of food retailers and advertisements were labeled as unhealthy. Most retailers were traditional corner stores (22%) and only 6% were fruit and vegetable markets. For advertisements, the prevailing food group promoted was carbonated and sugar-sweetened beverages (22%). The adjusted relative prevalence ratio of unhealthy to healthy food retailers was 1.9 times significantly higher in schools located in the richest areas (i.e., lowest poverty rates) as compared to the poorest areas (ARPR=1.9, 95% CI [1.3-2.7]). No significant associations were found between advertisements and school characteristics. Of the 2,465 children included in the study, 37% were overweight of whom 17% were obese. One quarter of children purchased foods from restaurants/ shops around schools. The adjusted multilevel models showed that the odds of being obese for children whose school neighborhoods had a medium or a high density of unhealthy food retailers was 1.6 times (95% CI [1.08; 2.41]) and 1.7 times (95% CI [1.05; 2.76]) that of children whose school neighborhoods had low density of unhealthy retailers. A closer exploration of the data revealed that a medium and a high density of all the subtypes of unhealthy retailers were all positively associated with obesity with statistical significance being reached only for two categories: Bakeries and pastries as well as fast-food restaurants. As for distance from schools to nearest unhealthy retailers, we found no significant pattern across the different types. Our sensitivity analyses showed that results were comparable when using different metrics to assess availability of food retailers around schools; but not when using different weight-related outcomes.

Conclusion Our study showed that school neighborhood food environments included predominantly unhealthy food retailers and advertisements; and correlated the availability of unhealthy food retailers with children’s obesity in this context. As the global discourse around promoting healthy food environments has recently amplified, this study provides high-quality evidence on the impact of food environments on childhood obesity, and contributes to reducing the related research gaps in LMICs. It also contributes evidence in support of calls for policies to promote healthy food options and limit unhealthy ones in school neighborhoods, in an attempt to curb the emerging childhood obesity epidemic, particularly in nutrition transition contexts.